[LTP] [PATCH v8 1/4] Refactor regen.sh script to generate syscalls

[Andrea Cervesato]

Hi Petr,

On 11/1/24 08:39, Petr Vorel wrote:
> Hi all,
>
>> Hi!
>>> +while IFS= read -r arch; do
>>> +	(
>>> +		echo
>>> +		case ${arch} in
>>> +		sparc64) echo "#if defined(__sparc__) && defined(__arch64__)" ;;
>>> +		sparc) echo "#if defined(__sparc__) && !defined(__arch64__)" ;;
>>> +		s390) echo "#if defined(__s390__) && !defined(__s390x__)" ;;
>>> +		mips64n32) echo "#if defined(__mips__) && defined(_ABIN32)" ;;
>>> +		mips64) echo "#if defined(__mips__) && defined(_ABI64)" ;;
>>> +		mipso32) echo "#if defined(__mips__) && defined(_ABIO32) && _MIPS_SZLONG == 32" ;;
> The original code was:
>
> -		mips_n32) echo "#if defined(__mips__) && defined(_ABIN32)" ;;
> -		mips_n64) echo "#if defined(__mips__) && defined(_ABI64)" ;;
> -		mips_o32) echo "#if defined(__mips__) && defined(_ABIO32) && _MIPS_SZLONG == 32" ;;
>
> ifdefs are the same, the only thing which changed are arch names. And these are
> also changed, see below. The only problem I see is that the change in
> supported-arch.txt below happens in the following commit. Shouldn't the rename
> happen in this commit already? syscalls.h file generated from the first commit
> is indeed different from the one generated in 2nd and 3rd (which are the same.
>
> +++ b/include/lapi/syscalls/supported-arch.txt
> @@ -1,13 +1,13 @@
> -aarch64
>   arc
> +arm64
>   arm
> -hppa
>   i386
>   ia64
> -loongarch
> -mips_n32
> -mips_n64
> -mips_o32
> +loongarch64
> +mips64n32
> +mips64
> +mipso32
> +parisc
>   powerpc64
>   powerpc
>   s390x
>
> 1st commit have (which is IMHO wrong)
> #ifdef __mips_n32__
>
> 2nd and 3rd have:
> #if defined(__mips__) && defined(_ABIN32)
>
> The reason due postponed changes is, that some arch use
>
> *) echo "#ifdef __${arch}__" ;;
>
> Whoever will do bisecting, cannot use this commit on mips and other affected
> archs due broken syscalls.h. This problem appeared in v5. Previously (v3 and v4)
> old names were used in generate_syscalls.sh, and in second commit (when the
> generate_arch.sh archs was added) were new ones used.
>
> IMHO clearest solution would be to move rename to 1st commit.
I agree with it. I'm going to fix it.
>
>> It would be nice to get a confirmation from someone who knows mips that
>> these changes in ifdefs are okay, otherwise:
> I guess we know the answer thus we don't have to bother mips folks.
> But very good catch.
>
> Kind regards,
> Petr
Andrea